Merge branch 'main' into pmu-data-overlay

Author: jmahbs

.ci/scripts/build-qnn-sdk.sh

@@ -18,7 +18,7 @@ build_qnn_backend() {
   export EXECUTORCH_ROOT="$(cd -- "$(dirname -- "${BASH_SOURCE[0]}")/../.." && pwd)"
 
   parallelism=$(( $(nproc) - 1 ))
-  bash backends/qualcomm/scripts/build.sh --skip_aarch64 --job_number ${parallelism} --release
+  bash backends/qualcomm/scripts/build.sh --skip_linux_android --skip_linux_embedded --job_number ${parallelism} --release
 }
 
 set_up_aot() {

.github/workflows/pull.yml

@@ -909,6 +909,8 @@ jobs:
 
   test-samsung-models-linux:
     name: test-samsung-models-linux
+    # Skip this job if the pull request is from a fork (secrets are not available)
+    if: github.event.pull_request.head.repo.full_name == github.repository || github.event_name != 'pull_request'
     uses: pytorch/test-infra/.github/workflows/linux_job_v2.yml@main
     permissions:
       id-token: write

backends/aoti/common_shims.cpp

@@ -184,6 +184,10 @@ int32_t aoti_torch_dtype_int32() {
   return 3; // PyTorch's int32 dtype code
 }
 
+int32_t aoti_torch_dtype_bool() {
+  return 11; // PyTorch's bool dtype code
+}
+
 int32_t aoti_torch_dtype_int64() {
   return 4; // PyTorch's int64 dtype code
 }

backends/aoti/common_shims.h

@@ -63,6 +63,7 @@ int32_t aoti_torch_dtype_int8();
 int32_t aoti_torch_dtype_int16();
 int32_t aoti_torch_dtype_int32();
 int32_t aoti_torch_dtype_int64();
+int32_t aoti_torch_dtype_bool();
 
 // Dtype utility function needed by Metal backend
 size_t aoti_torch_dtype_element_size(int32_t dtype);

backends/aoti/tests/test_common_shims.cpp

@@ -322,3 +322,14 @@ TEST_F(CommonShimsTest, IndependentCaches) {
   // Sizes and strides pointers should be different (different caches)
   EXPECT_NE(sizes_ptr1, strides_ptr1);
 }
+
+// Test all dtype functions return correct PyTorch dtype codes
+TEST_F(CommonShimsTest, AllDtypesReturnCorrectValues) {
+  EXPECT_EQ(aoti_torch_dtype_float32(), 6); // PyTorch's float32 dtype code
+  EXPECT_EQ(aoti_torch_dtype_bfloat16(), 15); // PyTorch's bfloat16 dtype code
+  EXPECT_EQ(aoti_torch_dtype_int8(), 1); // PyTorch's int8 dtype code
+  EXPECT_EQ(aoti_torch_dtype_int16(), 2); // PyTorch's int16 dtype code
+  EXPECT_EQ(aoti_torch_dtype_int32(), 3); // PyTorch's int32 dtype code
+  EXPECT_EQ(aoti_torch_dtype_int64(), 4); // PyTorch's int64 dtype code
+  EXPECT_EQ(aoti_torch_dtype_bool(), 11); // PyTorch's bool dtype code
+}

backends/aoti/utils.h

@@ -45,6 +45,8 @@ inline executorch::aten::ScalarType dtype_to_scalar_type(int32_t dtype) {
       return executorch::aten::ScalarType::Long;
     case 6: // PyTorch's float32 dtype code
       return executorch::aten::ScalarType::Float;
+    case 11: // PyTorch's bool dtype code
+      return executorch::aten::ScalarType::Bool;
     case 15: // PyTorch's bfloat16 dtype code
       return executorch::aten::ScalarType::BFloat16;
     // Future support for additional dtypes can be added here

backends/cuda/runtime/utils.h

@@ -62,6 +62,7 @@ enum class SupportedDTypes : int32_t {
   INT32 = 3, // PyTorch's int32 dtype code
   INT64 = 4, // PyTorch's int64 dtype code
   FLOAT32 = 6, // PyTorch's float32 dtype code
+  BOOL = 11, // PyTorch's bool dtype code
   BFLOAT16 = 15, // PyTorch's bfloat16 dtype code
 };
 
@@ -84,6 +85,7 @@ inline bool is_dtype_supported_in_et_cuda(int32_t dtype) {
     case static_cast<int32_t>(SupportedDTypes::INT32):
     case static_cast<int32_t>(SupportedDTypes::INT64):
     case static_cast<int32_t>(SupportedDTypes::FLOAT32):
+    case static_cast<int32_t>(SupportedDTypes::BOOL):
     case static_cast<int32_t>(SupportedDTypes::BFLOAT16):
       return true;
     default:
@@ -96,13 +98,14 @@ inline AOTITorchError validate_dtype(int32_t dtype) {
   ET_CHECK_OR_RETURN_ERROR(
       is_dtype_supported_in_et_cuda(dtype),
       InvalidArgument,
-      "Unsupported dtype: %d. Supported dtypes: %d (int8), %d (int16), %d (int32), %d (int64), %d (float32), %d (bfloat16)",
+      "Unsupported dtype: %d. Supported dtypes: %d (int8), %d (int16), %d (int32), %d (int64), %d (float32), %d (bool), %d (bfloat16)",
       dtype,
       static_cast<int32_t>(SupportedDTypes::INT8),
       static_cast<int32_t>(SupportedDTypes::INT16),
       static_cast<int32_t>(SupportedDTypes::INT32),
       static_cast<int32_t>(SupportedDTypes::INT64),
       static_cast<int32_t>(SupportedDTypes::FLOAT32),
+      static_cast<int32_t>(SupportedDTypes::BOOL),
       static_cast<int32_t>(SupportedDTypes::BFLOAT16));
 
   return Error::Ok;

backends/qualcomm/README.md

@@ -27,6 +27,7 @@ Please check `generate_qnn_executorch_compiler_spec()` in
 - SXR1230P
 - SXR2230P
 - SXR2330P
+- QCS9100
 
 ### Adding more supported Chipset
 Currently, users cannot add additional chipset models because the chipset ID is not accessible to community users. If you have specific chipset models you wish to add, please contact one of the authors in the `Code Reviews` section at the bottom of this page.

backends/qualcomm/debugger/README.md

@@ -54,6 +54,7 @@ adb = SimpleADB(
     device_id=args.device,
     host_id=args.host,
     soc_model=args.model,
+    target=args.target,
 )
 binaries_trace = generate_optrace(
     args, adb, f"{args.artifact}/{pte_filename}.pte", example_input

backends/qualcomm/scripts/build.sh

@@ -25,8 +25,10 @@ set -o xtrace
 usage() {
   echo "Usage: Build the aarch64 version of executor runner or the python interface of Qnn Manager"
   echo "First, you need to set the environment variable for QNN_SDK_ROOT"
-  echo ", and if you want to build the aarch64 version of executor runner"
+  echo ", and if you want to build the android version of executor runner"
   echo ", you need to export ANDROID_NDK_ROOT=/path/to/android_ndkXX"
+  echo "(or export TOOLCHAIN_ROOT_HOST=/path/to/sysroots/xx_host, "
+  echo "TOOLCHAIN_ROOT_TARGET=/path/to/sysroots/xx_target for linux embedded with --enable_linux_embedded)"
   echo "e.g.: executorch$ ./backends/qualcomm/scripts/build.sh --skip_x86_64"
   exit 1
 }
@@ -36,8 +38,10 @@ usage() {
 
 BUILD_X86_64="true"
 CMAKE_X86_64="build-x86"
-BUILD_AARCH64="true"
-CMAKE_AARCH64="build-android"
+BUILD_ANDROID="true"
+CMAKE_ANDROID="build-android"
+BUILD_OE_LINUX="false"
+CMAKE_OE_LINUX="build-oe-linux"
 CLEAN="true"
 BUILD_TYPE="RelWithDebInfo"
 BUILD_JOB_NUMBER="16"
@@ -50,7 +54,7 @@ if [ -z BUCK2 ]; then
   BUCK2="buck2"
 fi
 
-long_options=skip_x86_64,skip_aarch64,no_clean,release,job_number:
+long_options=skip_x86_64,skip_linux_android,skip_linux_embedded,enable_linux_embedded,no_clean,release,job_number:
 
 parsed_args=$(getopt -a --options '' --longoptions $long_options --name "$0" -- "$@")
 eval set -- "$parsed_args"
@@ -59,7 +63,9 @@ eval set -- "$parsed_args"
 while true ; do
     case "$1" in
         --skip_x86_64) BUILD_X86_64="false"; shift;;
-        --skip_aarch64) BUILD_AARCH64="false"; shift;;
+        --skip_linux_android) BUILD_ANDROID="false"; shift;;
+        --skip_linux_embedded) BUILD_OE_LINUX="false"; shift;;
+        --enable_linux_embedded) BUILD_ANDROID="false"; BUILD_OE_LINUX="true"; shift;;
         --no_clean) CLEAN="false"; shift;;
         --release) BUILD_TYPE="Release"; shift;;
         --job_number) BUILD_JOB_NUMBER="$2"; shift 2;;
@@ -69,13 +75,13 @@ done
 
 PRJ_ROOT="$( cd "$(dirname "$0")/../../.." ; pwd -P)"
 
-if [ "$BUILD_AARCH64" = true ]; then
+if [ "$BUILD_ANDROID" = true ]; then
     if [[ -z ${ANDROID_NDK_ROOT} ]]; then
         echo "Please export ANDROID_NDK_ROOT=/path/to/android_ndkXX"
         exit -1
     fi
 
-    BUILD_ROOT=$PRJ_ROOT/$CMAKE_AARCH64
+    BUILD_ROOT=$PRJ_ROOT/$CMAKE_ANDROID
     if [ "$CLEAN" = true ]; then
         rm -rf $BUILD_ROOT && mkdir $BUILD_ROOT
     else
@@ -143,6 +149,94 @@ if [ "$BUILD_AARCH64" = true ]; then
     cmake --build $LLAMA_EXAMPLE_ROOT -j$BUILD_JOB_NUMBER
 fi
 
+if [ "$BUILD_OE_LINUX" = true ]; then
+    if [[ -z ${TOOLCHAIN_ROOT_HOST} ]]; then
+        echo "Please export e.g. TOOLCHAIN_ROOT_HOST=/path/to/sysroots/x86_64-qtisdk-linux"
+        exit -1
+    fi
+    if [[ -z ${TOOLCHAIN_ROOT_TARGET} ]]; then
+        echo "Please export e.g. TOOLCHAIN_ROOT_TARGET=/path/to/sysroots/armv8a-oe-linux"
+        exit -1
+    fi
+
+    BUILD_ROOT=$PRJ_ROOT/$CMAKE_OE_LINUX
+    if [ "$CLEAN" = true ]; then
+        rm -rf $BUILD_ROOT && mkdir $BUILD_ROOT
+    else
+        # Force rebuild flatccrt for the correct platform
+        cd $BUILD_ROOT/third-party/flatcc && make clean
+    fi
+
+    TOOLCHAN_PREFIX=$TOOLCHAIN_ROOT_HOST/usr/bin/aarch64-oe-linux/aarch64-oe-linux-
+    cd $BUILD_ROOT
+    cmake .. \
+        -DCMAKE_INSTALL_PREFIX=$BUILD_ROOT \
+        -DCMAKE_BUILD_TYPE=$BUILD_TYPE \
+        -DCMAKE_C_COMPILER=${TOOLCHAN_PREFIX}gcc \
+        -DCMAKE_CXX_COMPILER=${TOOLCHAN_PREFIX}g++ \
+        -DCMAKE_SYSROOT=$TOOLCHAIN_ROOT_TARGET \
+        -DCMAKE_SYSTEM_NAME=Linux \
+        -DCMAKE_SYSTEM_PROCESSOR=aarch64 \
+        -DCMAKE_FIND_ROOT_PATH_MODE_PROGRAM=NEVER \
+        -DEXECUTORCH_BUILD_QNN=ON \
+        -DEXECUTORCH_BUILD_DEVTOOLS=ON \
+        -DEXECUTORCH_BUILD_EXTENSION_LLM=ON \
+        -DEXECUTORCH_BUILD_EXTENSION_LLM_RUNNER=ON \
+        -DEXECUTORCH_BUILD_EXTENSION_MODULE=ON \
+        -DEXECUTORCH_BUILD_EXTENSION_DATA_LOADER=ON \
+        -DEXECUTORCH_BUILD_EXTENSION_FLAT_TENSOR=ON \
+        -DEXECUTORCH_BUILD_EXTENSION_NAMED_DATA_MAP=ON \
+        -DEXECUTORCH_BUILD_EXTENSION_TENSOR=ON \
+        -DEXECUTORCH_ENABLE_EVENT_TRACER=ON \
+        -DEXECUTORCH_ENABLE_LOGGING=ON \
+        -DQNN_SDK_ROOT=$QNN_SDK_ROOT \
+        -DEXECUTORCH_BUILD_KERNELS_QUANTIZED=ON \
+        -DPYTHON_EXECUTABLE=$PYTHON_EXECUTABLE \
+        -B$BUILD_ROOT
+
+    cmake --build $BUILD_ROOT -j$BUILD_JOB_NUMBER --target install
+
+    EXAMPLE_ROOT=examples/qualcomm
+    CMAKE_PREFIX_PATH="${BUILD_ROOT};${BUILD_ROOT}/third-party/gflags;"
+
+    cmake $PRJ_ROOT/$EXAMPLE_ROOT \
+        -DCMAKE_BUILD_TYPE=$BUILD_TYPE \
+        -DCMAKE_PREFIX_PATH=$CMAKE_PREFIX_PATH \
+        -DSUPPORT_REGEX_LOOKAHEAD=ON \
+        -DBUILD_TESTING=OFF \
+        -DEXECUTORCH_ENABLE_LOGGING=ON \
+        -DCMAKE_C_COMPILER=${TOOLCHAN_PREFIX}gcc \
+        -DCMAKE_CXX_COMPILER=${TOOLCHAN_PREFIX}g++ \
+        -DCMAKE_SYSROOT=$TOOLCHAIN_ROOT_TARGET \
+        -DCMAKE_SYSTEM_NAME=Linux \
+        -DCMAKE_SYSTEM_PROCESSOR=aarch64 \
+        -DEXECUTORCH_BUILD_KERNELS_QUANTIZED=ON \
+        -DCMAKE_FIND_ROOT_PATH_MODE_PROGRAM=NEVER \
+        -DCMAKE_FIND_ROOT_PATH_MODE_PACKAGE=BOTH \
+        -DPYTHON_EXECUTABLE=$PYTHON_EXECUTABLE \
+        -B$EXAMPLE_ROOT
+
+    cmake --build $EXAMPLE_ROOT -j$BUILD_JOB_NUMBER
+
+    LLAMA_EXAMPLE_ROOT=examples/models/llama
+    cmake $PRJ_ROOT/$LLAMA_EXAMPLE_ROOT \
+        -DBUILD_TESTING=OFF \
+        -DCMAKE_BUILD_TYPE=$BUILD_TYPE \
+        -DCMAKE_C_COMPILER=${TOOLCHAN_PREFIX}gcc \
+        -DCMAKE_CXX_COMPILER=${TOOLCHAN_PREFIX}g++ \
+        -DCMAKE_SYSROOT=$TOOLCHAIN_ROOT_TARGET \
+        -DCMAKE_SYSTEM_NAME=Linux \
+        -DCMAKE_SYSTEM_PROCESSOR=aarch64 \
+        -DCMAKE_FIND_ROOT_PATH_MODE_PROGRAM=NEVER \
+        -DCMAKE_PREFIX_PATH=$CMAKE_PREFIX_PATH \
+        -DEXECUTORCH_ENABLE_LOGGING=ON \
+        -DCMAKE_FIND_ROOT_PATH_MODE_PACKAGE=BOTH \
+        -DPYTHON_EXECUTABLE=$PYTHON_EXECUTABLE \
+        -B$LLAMA_EXAMPLE_ROOT
+
+    cmake --build $LLAMA_EXAMPLE_ROOT -j$BUILD_JOB_NUMBER
+fi
+
 if [ "$BUILD_X86_64" = true ]; then
     BUILD_ROOT=$PRJ_ROOT/$CMAKE_X86_64
     if [ "$CLEAN" = true ]; then